MeLOn
example_training_of_GP.m File Reference

Typedefs

using X = lhsdesign(nX,DX)
 

Functions

file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University Xiaopeng Daniel and Alexander Mitsos date January clc clear all close all addpath ("Direct")
 
Add path for DIRECT solver addpath ("GP training")
 
Add path for Gaussian process training functions addpath ("Write GP to files")
 
Evaluate test_func for all X Tranining of GPs Opt GP (1).matern
 
id GP_Parameters ()
 
 Write_GP_to_json (join([filename, ".json"]), Opt.GP(1), X, Y, lb, ub)
 
Compute GP predictions in Matlab (just for information) x_Test_Point
 
Plot generated data (just for information) % Generate a mesh on the inputs[x_1_prediction
 
 y_prediction (i, j)
 
 y_std (i, j)
 

Variables

file example_training_of_GP m brief Illustrative training script for Gaussian processes in Matlab
 
file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University n
 
file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University Xiaopeng Lin
 
file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University Xiaopeng Daniel Grothe
 
Add path for Gaussian process export functions General nX = 20
 
Number of training data points DX = 2
 
Input dimension of data GP lb = [-3, -3]
 
Define Lower bound of inputs ub = [ 3, 3]
 
Define upper bound of inputs test_func
 
Scale inputs onto interval[lb, ubY = cellfun(test_func, num2cell(X,2))
 
Define covariance function Martern
 
internal option for solver Opt = Train_GP_and_return_hyperparameters(X,Y,lb,ub,Opt)
 
Plot generated x_2_prediction = meshgrid(linspace(lb(1), ub(1), 100), linspace(lb(2), ub(2),100) )
 
 y_prediction = zeros(size(x_2_prediction,1),1)
 
 y_std = zeros(size(x_2_prediction,1),1)
 
Evaluate prediction at all mesh points for i
 

Typedef Documentation

◆ X

Function for data generation Generate training data X = lhsdesign(nX,DX)

Function Documentation

◆ addpath() [1/3]

file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University Xiaopeng Daniel and Alexander Mitsos date January clc clear all close all addpath ( "Direct"  )

◆ addpath() [2/3]

Add path for DIRECT solver addpath ( "GP training )

◆ addpath() [3/3]

Add path for Gaussian process training functions addpath ( "Write GP to files"  )

◆ data()

Plot generated data ( just for  information)

◆ GP()

Evaluate test_func for all X Tranining of GPs Opt GP ( )

◆ GP_Parameters()

id GP_Parameters ( )
virtual

◆ Matlab()

Compute GP predictions in Matlab ( just for  information)

◆ Write_GP_to_json()

Write_GP_to_json ( join([filename, ".json"])  ,
Opt.  GP1,
X  ,
Y  ,
lb  ,
ub   
)

◆ y_prediction()

y_prediction ( i  ,
j   
)

◆ y_std()

y_std ( i  ,
j   
)

Variable Documentation

◆ DX

Number of training data points DX = 2

◆ Grothe

file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University Xiaopeng Daniel Grothe

◆ i

Evaluate prediction at all mesh points for i
Initial value:
= 1 : size(x_1_prediction,1)
for j = 1 : size(x_2_prediction,1)
x_Test_Point = [x_1_prediction(i,j), x_2_prediction(i,j)]

◆ lb

Input dimension of data GP lb = [-3, -3]

◆ Lin

file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University Xiaopeng Lin

◆ Martern

Define covariance function Martern

◆ Matlab

file example_training_of_GP m brief Illustrative training script for Gaussian processes in Matlab
Initial value:
==============================================================================\n
% Aachener Verfahrenstechnik-Systemverfahrenstechnik

◆ n

file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University n
Initial value:
==============================================================================\n
%
% @author Artur Schweidtmann

◆ nX

Add path for Gaussian process export functions General nX = 20

◆ Opt

internal option for solver Opt = Train_GP_and_return_hyperparameters(X,Y,lb,ub,Opt)

◆ test_func

Define upper bound of inputs test_func
Initial value:
= @(x) 3*(1-x(1)).^2.*exp(-(x(1).^2) - (x(2)+1).^2) ...
- 10*(x(1)/5 - x(1).^3 - x(2).^5).*exp(-x(1).^2-x(2).^2) ...
- 1/3*exp(-(x(1)+1).^2 - x(2).^2)

◆ ub

Define Lower bound of inputs ub = [ 3, 3]

◆ x_2_prediction

Plot generated x_2_prediction = meshgrid(linspace(lb(1), ub(1), 100), linspace(lb(2), ub(2),100) )

◆ Y

Scale inputs onto interval [lb, ub] Y = cellfun(test_func, num2cell(X,2))

◆ y_prediction

y_prediction = zeros(size(x_2_prediction,1),1)

◆ y_std

y_std = zeros(size(x_2_prediction,1),1)
x_2_prediction
Plot generated x_2_prediction
Definition: example_training_of_GP.m:65
size
Xnew size()
i
Evaluate prediction at all mesh points for i
Definition: example_training_of_GP.m:72
n
file example_training_of_GP m brief Illustrative training script for Gaussian processes in RWTH Aachen University n
Definition: example_training_of_GP.m:8
j
Training of GP for j
Definition: Train_GP_and_return_hyperparameters.m:24